rt: Stop using atomic ops on rust_kernel::live_tasks

These ops are all done within spitting distance of a suitable lock,
so just protect it with the lock.

Author: brson

src/rt/rust_kernel.cpp

@@ -84,25 +84,27 @@ rust_kernel::fail() {
 
 void
 rust_kernel::register_task(rust_task *task) {
+    int new_live_tasks;
     {
         scoped_lock with(task_lock);
         task->user.id = max_task_id++;
         task_table.put(task->user.id, task);
+        new_live_tasks = ++live_tasks;
     }
     K(srv, task->user.id != INTPTR_MAX, "Hit the maximum task id");
     KLOG_("Registered task %" PRIdPTR, task->user.id);
-    int new_live_tasks = sync::increment(live_tasks);
     KLOG_("Total outstanding tasks: %d", new_live_tasks);
 }
 
 void
 rust_kernel::release_task_id(rust_task_id id) {
     KLOG_("Releasing task %" PRIdPTR, id);
+    int new_live_tasks;
     {
         scoped_lock with(task_lock);
         task_table.remove(id);
+        new_live_tasks = --live_tasks;
     }
-    int new_live_tasks = sync::decrement(live_tasks);
     KLOG_("Total outstanding tasks: %d", new_live_tasks);
     if (new_live_tasks == 0) {
         // There are no more tasks and there never will be.

src/rt/rust_kernel.h

@@ -22,12 +22,13 @@ class rust_kernel {
 private:
     rust_scheduler *sched;
 
+    // Protects live_tasks, max_task_id and task_table
+    lock_and_signal task_lock;
     // Tracks the number of tasks that are being managed by
     // schedulers. When this hits 0 we will tell all schedulers
     // to exit.
-    volatile int live_tasks;
-    // Protects max_task_id and task_table
-    lock_and_signal task_lock;
+    int live_tasks;
+    // The next task id
     rust_task_id max_task_id;
     hash_map<rust_task_id, rust_task *> task_table;